Isles of Scilly Project
The Isles of Scilly, an archipelago 25 miles off the south-west tip of Cornwall, are one of the few places in the UK that are thought to be free from Varroa. The islands accommodate a small number of honey bee colonies, predominantly of the dark variety, Apis mellifera mellifera, with imports in recent years known to have been made from the Isle of Man and Colonsay, as well as some of ‘Buckfast’ origin from the mainland, and possibly some exotic imports from further afield 1. Inter-island movement of bees, whether naturally or by human intervention, will have implications for bee genetics and disease transfer. The aim of this project is to characterise the present-day genetic diversity of managed honey bees on the islands, and to revisit this in future years to establish the impact of bee movements on their genetics.
The network below illustrates the results from sampling in 2021. Each pie chart node represents a single drone, and the node outline colour indicates the island from which it was sampled. The pie charts present the genetic admixture proportions for the C (yellow), M (black), and O (blue) lineages, and the actual values can be viewed by hovering the mouse over a given node. Lines (edges) connecting nodes represent the most genetically similar drone pairs (up to 5 per drone), and the edge thickness relates to the drone pair’s identity by state (IBS) score. IBS is on a scale of 0 (completely different) to 1 (identical) and the value for a given edge can be viewed by hovering the mouse over the edge. The network can be dragged and zoomed.
These preliminary results imply shared genetic ancestry between drones sampled from Bryher and Tresco. This might arise due to a number of possible scenarios, including: human-mediated activities such as beekeepers on different islands importing from the same breeder, or a beekeeper on one island sourcing bees from the other; or naturally if for example drones or queens “island-hop” to mate at a congregation area, or if a swarm migrates.
The technical stuff
llumina whole-genome sequencing data (at 10 X depth of coverage) for 55 Scillonian bees sampled from 11 apiaries was aligned to the honey bee reference genome (Amel_HAv3.1; https://www.ncbi.nlm.nih.gov/assembly/GCF_003254395.2). Variants identified with HaplotypeCaller were joint-called across all samples, resulting in 3.5M single-nucleotide polymorphisms (SNPs) with a mean call rate of 97%. Missing genotypes were imputed with BEAGLE (v5.4) using the AmelHap reference panel (https://doi.org/10.5281/zenodo.6983103) comprising more than 1300 drone haplotypes. llumina whole-genome sequencing data for 139 honey bee workers, generated for the SMART bees project (https://doi.org/10.3390/genes13020182), were also aligned to Amel_HAv3.1, and genotyped at the AmelHap variants. The data was phased and missing genotypes imputed with BEAGLE (v5.4) again using the AmelHap reference panel. Both datasets were merged and filtered to: (1) remove variants present in one dataset but not the other; and (2) remove variants in strong linkage disequilibrium (r2 > 0.1 for pairs of SNPs up to 10 kb apart). The final dataset comprised 919,951 SNPs and 194 bees. An identity by state (IBS) matrix was generated using Plink. Local ancestry estimates for k = 3 genetic backgrounds were calculated with alstructure in R by analysing each Scillonian bee independently with the SMART bees, following which the SMART bee ancestry estimates were averaged across all runs and merged with the Scillonian bee ancestry estimates. The network diagram was generated in R from the IBS matrix. For each drone the top five IBS pairs were retained, and these were then filtered to remove drone pairs with IBS < 0.8. The network was generated with the visNetwork package, and pie chart images of the alstructure admixture estimates plotted on the nodes.
Analyses are ongoing and results are subject to change.